samba.git/source4/torture/raw, branch master

torture: Fix raw.open.chained-ntcreatex

2014-12-09T18:43:31+00:00

sizeof(buf) is 5. On FreeBSD10/clang this overwrites "ret". Not good.

Same as 574750777a

Signed-off-by: Volker Lendecke 
Reviewed-by: Jeremy Allison 

Autobuild-User(master): Jeremy Allison 
Autobuild-Date(master): Tue Dec  9 19:43:31 CET 2014 on sn-devel-104

torture: Fix raw.open.chained-openx

2014-11-28T12:30:18+00:00

sizeof(buf) is 5. On FreeBSD10/clang this overwrites "ret". Not good.

Signed-off-by: Volker Lendecke 
Reviewed-by: Ralph Boehme 

Autobuild-User(master): Ralph Böhme 
Autobuild-Date(master): Fri Nov 28 13:30:18 CET 2014 on sn-devel-104

s4:torture: Add smb2.oplock test batch9a and raw.oplock test batch9a

2014-11-07T19:13:09+00:00

Shows attribute(stat) access open can create a file,
and subsequent attribute(stat) opens don't break oplocks.

Can be extended to explore more varients.

Signed-off-by: Jeremy Allison 
Reviewed-by: Stefan Metzmacher

torture: Provide enough space for test EA name in raw.eas test

2014-09-11T06:50:16+00:00

The issue is that previously bad_ea_name[5] was the last element on
the array, and so when we later did a strlen() on it, we read past the
end of the stack array.  We need bad_ea_name[5] to be the second-last
element, followed by the \0 placed there by the strlcpy().

Found by AddressSanitizer

Change-Id: I871c08200aa2591c612dfa44da92b83132f83d88
Signed-off-by: Andrew Bartlett 
Reviewed-by: Kamen Mazdrashki 

Autobuild-User(master): Andrew Bartlett 
Autobuild-Date(master): Thu Sep 11 08:50:16 CEST 2014 on sn-devel-104

torture: allow us to correctly use sizeof(buf) in raw.open tests

2014-09-08T05:26:34+00:00

This changes the sizeof(buf) from sizeof(void *), 8 on 64-bit machines, to sizeof("test") (eg 5).

Found by AddressSanitizer

Andrew Bartlett

Change-Id: I01f18b35c041f3b16be9f6da8ae5d1917d7e24d9
Signed-off-by: Andrew Bartlett 
Reviewed-by: Kamen Mazdrashki

torture: Use torture_assert macro for value check in raw.read

2014-08-29T22:27:13+00:00

Signed-off-by: Christof Schmitt 
Reviewed-by: Volker Lendecke

torture: Use torture_assert macro for status check in raw.read

2014-08-29T22:27:13+00:00

Signed-off-by: Christof Schmitt 
Reviewed-by: Volker Lendecke

torture: Use torture_fail macro in check_buffer for read requests

2014-08-29T22:27:13+00:00

Signed-off-by: Christof Schmitt 
Reviewed-by: Volker Lendecke

torture: Add test for 16 bit alignment of readx data

2014-08-29T22:27:13+00:00

MS-CIFS requires a one byte pad to guarantee 16 bit alignment of the
data:

Pad (1 byte): This field is optional. When using the NT LAN Manager
dialect, this field can be used to align the Data field to a 16-bit
boundary relative to the start of the SMB Header. If Unicode strings are
being used, this field MUST be present. When used, this field MUST be
one padding byte long.

Signed-off-by: Christof Schmitt 
Reviewed-by: Volker Lendecke

s4: torture: Add a new lock test to show that the Samba SMB1 multi-lock implementation is (currently) correct.

2014-07-08T08:16:59+00:00

Needed as there was a proposal to re-architect
our multi-lock to dispense with lock order precedence,
which isn't how Windows does it (unfortunately,
as the new code would have been cleaner :-).
Tested against the Win2k12 SMB1 implementation.

This test is designed to show that
lock precedence on the server is based
on the order received, not on the ability
to grant. For example:

A blocked lock request containing 2 locks
will be satified before a subsequent blocked
lock request over one of the same regions,
even if that region is then unlocked. E.g.

(a) lock 100->109, 120->129 (granted)
(b) lock 100->109, 120-129 (blocks)
(c) lock 100->109          (blocks)
(d) unlock 100->109

lock (c) will not be granted as lock (b)
will take precedence.

Signed-off-by: Jeremy Allison 
Reviewed-by: Andreas Schneider 

Autobuild-User(master): Andreas Schneider 
Autobuild-Date(master): Tue Jul  8 10:16:59 CEST 2014 on sn-devel-104